Universal Basic Income Works — Just Not on the Problem AI Creates

An episode of Dan's AI Intel

Cash reliably cures poverty and doesn't make people lazy — but it can't manufacture the one thing a post-work economy takes away: a reason to get up in the morning. The most-studied idea for life after work succeeds at the problem we don't have yet and fails at the one AI may create — which is why its own funders are trading income for ownership.

Published · Updated · By Dan Walter

Executive summary

The most important thing the universal-basic-income experiments have taught us is not whether the idea works but which half of it works — and the answer should unsettle everyone in the AI debate. Across fifty years and four continents, unconditional cash has reliably done the thing it was designed to do: it reduces poverty, improves health, keeps children in school, and, contrary to the oldest objection against it, does not make people stop working. By the standard of an anti-poverty tool, basic income is one of the better-evidenced ideas in social policy. Yet it is failing at the one task the AI age would actually set it. A cheque restores income; it cannot restore the structure, purpose, status and belonging that a job supplies as freight — the very things a genuinely post-work economy threatens to strip away. That gap, not cost, is the deepest finding here, and over the long run it matters more than any single pilot result.

It is also why the people who built the experiment are abandoning their own conclusion. At least half a dozen of the field's most prominent voices — Musk, Yang, Zuckerberg, Hinton among them — have called a basic income unavoidable in an age of intelligent machines. But Sam Altman, who funded the largest unconditional-cash study in American history, reversed himself in public the moment its results landed: a fixed cheque, he now says, "does not get [us] what we're really going to need." His pivot, and the wider drift from "income" to "ownership" — equity, "universal basic compute," public wealth funds — is the tell. It is a more radical proposal, because it changes who owns the machines rather than who receives a cheque drawn on them, and a more self-serving one, advanced largely by the people who own those machines today. But a dividend is still a cheque. It answers the ownership question and walks straight past the meaning one.

So the forward question is not "how big should the payment be?" The evidence quietly demolishes the comfortable certainties on both sides: the pilots prove the cheque does not breed laziness, and they also prove — through their own design — that a temporary top-up to a lucky few tells us almost nothing about a permanent, universal floor for everyone, which would collide with costs no electorate has accepted and general-equilibrium effects no experiment can measure. The real hinge is anthropological. If the labs are wrong about AI, basic income stays a useful, partial anti-poverty measure and no revolution. If they are right, a civilisation that can automate cognition can probably afford to feed everyone — and will discover that the binding constraint was never money but meaning, the one thing the experiments suggest no cheque has ever managed to buy.

Why a fight about welfare is really a fight about the next decade

For two hundred years, the idea of simply handing everyone money was a curiosity at the edge of economics — argued over by a few philosophers, tried in a handful of towns, and never adopted anywhere at full strength. Then the people building artificial intelligence picked it up, and it stopped being a curiosity. When the engineers who expect their own machines to perform a large share of human labour start saying that a no-strings income may be the only humane way through, the question changes character. It is no longer "should the welfare state be more generous?" It is "if work itself becomes optional for tens of millions of people, what holds a society together?"

That is the real subject here. The defining claim of this moment — made by the labs themselves, not by their critics — is that within years, not generations, software will do cognitive work that today requires a salaried human. If that claim is even half right, the twentieth-century deal in which most people trade labour for income comes under strain, and universal basic income is the most developed answer anyone has offered to it. So the case for and against it is not a side-quest in welfare policy; it is a direct test of whether the future these companies are building is one a person could actually live in.

The good news is that we are not arguing in the dark. Unlike almost every other question about the AI economy, this one has been studied for half a century, on four continents, with real money and real randomised controls. The evidence is unusually rich, and it points somewhere genuinely surprising — somewhere that should unsettle both the people selling basic income and the people sneering at it.

The headcount: who signed the cheque, and who is quietly tearing it up

Start with the roll call, because it is more impressive than the sceptics admit and more fragile than the believers like to mention.

Elon Musk has said for years that advanced AI and robotics will eliminate the need for most jobs, and that some form of guaranteed income becomes unavoidable as a result — he prefers to call it "universal high income" rather than basic, and paints it as the on-ramp to a "Star Trek future" of post-scarcity abundance. Andrew Yang built an entire 2020 presidential campaign on a $1,000-a-month "Freedom Dividend," arguing flatly that "AI will wind up funding universal income" and that the only question was how fast. Mark Zuckerberg used his 2017 Harvard commencement address to tell graduates the country should "explore ideas like universal basic income." Geoffrey Hinton — the computer scientist whose work underpins the deep-learning boom and who left Google in 2023 to warn about where it leads — went further than most, saying he had been consulted by the UK government and had told them a basic income was a sensible response to the job losses he expects. Facebook's co-founder Chris Hughes put real money into guaranteed-income research. At least half a dozen of the most prominent figures in and around the field have, at some point, called a basic income not just desirable but inevitable.

And then there is Sam Altman, who did something rarer than endorse the idea: he tested it. Through a research outfit called OpenResearch, the OpenAI chief executive backed what became the largest unconditional-cash study in modern American history — and then, just as its results landed, he changed his mind in public. "I no longer believe in universal basic income as much as I once did," he said in a 2025 interview. A fixed monthly cheque, he argued, "does not get [us] what we're really going to need for this next phase." His attention has moved from income to ownership — equity stakes in the machines, "universal basic compute" that would give every person a slice of AI's raw capability, even a half-serious proposal to measure the world's AI output in tokens and parcel a share of it out to all eight billion humans.

That reversal is the first crack in the consensus, and it is not the only one. The Nobel laureate Paul Krugman has been blunt: "I'm not a UBI guy," he says, because a basic income generous enough to matter would cost more than seems politically conceivable for decades, while a version cheap enough to pass would be too small to live on. The International Labour Organization has published a paper arguing in plain terms that a basic income is "not the answer" to AI's impact on work. And a whole school of thought — call it the adaptation camp — points out that every previous technological revolution destroyed particular jobs while creating more in aggregate, and that planning to pay everyone off assumes a helplessness in ordinary workers that history has never borne out.

So the honest tally is this: the headcount of believers is real, and it includes some of the smartest, best-informed people alive. But their conviction is fracturing exactly as the data arrives, and the most committed funder of the experiment is the one walking away fastest. That is a strange thing for a winning idea to do. To understand why, you have to be precise about what "basic income" even means — because half the public argument is two camps using one phrase for two different things.

What "basic" actually means — and why the word does so much work

A universal basic income, in its textbook form, has four properties at once: it is universal (everyone gets it, rich and poor alike), unconditional (no work requirement, no means test, no behaviour to prove), individual (paid to people, not households), and periodic (a regular cash payment, not a one-off grant or a voucher you can only spend on approved things). Strip away any of those four and you have something else — a targeted benefit, a tax credit, a one-time stake. The purity of the definition is the whole point: the promise is a floor below which no one can fall, that nothing you do can take away.

The word that does the most quiet work is "basic." Basic is a claim about scale: an income at the level needed to cover basic needs — roughly subsistence, somewhere around a national poverty line. By that strict standard, almost none of the experiments people cite as "UBI trials" tested a basic income at all. They tested partial basic incomes — $500 a month in Stockton, $1,000 in the OpenResearch study, €1,200 in Germany — sums that supplement a life rather than replace a wage. A genuinely full UBI in the United States, one a person could actually live on without working, would mean something on the order of twelve to fifteen thousand dollars per adult per year, and far more for a family. The gap between the partial pilots and the full promise is not a technicality. It is most of the reason the evidence is easier to read than the politics: we have strong data on what a top-up does and almost none on what a true, permanent, replace-your-wage floor would do.

Exhibit — Every 'UBI trial' tested a small, temporary top-up — never the real thing. A scattered sample paid for two or three years cannot tell us what a permanent floor for everyone would do. Source: OpenResearch · DIW Berlin · Kela · SEED (2019–21); Hoynes & Rothstein (NBER) on the full floor. Compiled by Dan's AI Intel.

Then there is the family of near-relations that get tangled together. A negative income tax, the version Milton Friedman proposed in 1962, delivers the floor through the tax system: instead of paying tax, people below a threshold receive a payment that is a fraction of how far their income falls short. A guaranteed or minimum income is the means-tested welfare model most countries already run — a floor, but one that claws back as you earn, sometimes at close to a dollar lost for every dollar gained, which creates the notorious "welfare trap" where taking a job barely pays. A job guarantee, favoured on the left, is the opposite philosophy entirely: not cash for nothing, but a standing offer of paid public work to anyone who wants it, with government as the employer of last resort.

Here is the subtlety that trips up most conversations. Friedman himself pointed out that a universal basic income and a negative income tax are, mathematically, almost the same thing: a basic income paid to everyone and then taxed back from those who don't need it produces the same final distribution as a negative income tax that only ever pays the people below the line. The economist Philippe Van Parijs, the idea's most rigorous modern defender, agreed they were economically equivalent — but insisted the timing changes behaviour. A basic income arrives before you act, every month, unconditionally; you hold the floor in your hand and plan around it. A negative income tax arrives after, at filing, as a reconciliation. The first gives you security you can feel and use; the second gives you the same money with none of the psychological permission. That distinction — security held in advance versus compensation paid in arrears — turns out to matter enormously when you look at what recipients actually do.

So when two people argue about "basic income," one is often defending an unconditional cheque to everyone and the other is attacking a means-tested handout to the poor, and they never notice they are describing different machines. Keep the four properties in mind, keep "partial versus full" in mind, and most of the noise resolves into a real and narrow disagreement.

The two-hundred-year argument: why the right and the left both want this

The strangest fact about basic income is that its most committed champions have spent two centuries agreeing on the policy while disagreeing about everything else. It is not a left idea or a right idea. It is an idea that the left and the right keep arriving at from opposite directions, for opposite reasons — which is its great political strength and, as we will see, its fatal political weakness.

The first recognisable version came from Thomas Paine, the pamphleteer of the American and French revolutions, in a 1797 essay called Agrarian Justice. Paine's argument was not about charity or efficiency; it was about property and theft. In the state of nature, he reasoned, the earth was the common inheritance of everyone. The moment land became private property, the people left without any became worse off than they would have been in the wild — robbed of a birthright they never consented to give up. The owners therefore owed the dispossessed not a kindness but a debt. Paine proposed paying it as a lump sum to every person reaching twenty-one — fifteen pounds, something like two and a half thousand of today's pounds — plus an annual pension from the age of fifty. It was, in embryo, a basic income justified entirely as a right: compensation for an enclosure none of us agreed to.

A century and a half later the idea was picked up by the unlikeliest of patrons. Milton Friedman, the patron saint of free markets, proposed his negative income tax not despite his conservatism but because of it. To Friedman, the sprawling apparatus of the welfare state — its dozens of programmes, its armies of caseworkers, its rules about what the poor could buy and how they had to behave — was an offence against both efficiency and freedom. Replace the whole machine with a single cash transfer, he argued, and you would help people more, cost less, and stop treating adults like wards of the state. The right-wing case for handing out money is that it is the least paternalistic thing a government can do: give people the cash and let them, not a bureaucrat, decide what they need.

And then, in 1995, the Belgian philosopher Philippe Van Parijs gave the idea its most ambitious modern grounding in a book titled Real Freedom for All. His move was to redefine freedom itself. Formal freedom — the right to do something the law permits — is worthless, he argued, if you lack the means to act on it. A person free to start a business but with no savings, free to leave an abusive job but with rent due Friday, free to write a novel but unable to eat, is not really free at all. Real freedom is the effective capacity to live as you choose, and the unconditional grant is the institution that delivers it to everyone, not just the propertied. Van Parijs cheerfully accepted the most provocative implication: that even a surfer choosing to ride waves at Malibu rather than work would be owed his share. That was the line that drew fire — the philosopher John Rawls among others objected that a society cannot ask the diligent to subsidise the voluntarily idle without violating some basic principle of reciprocity — and we will come back to it, because the reciprocity objection turns out to be the hinge of the entire debate.

Hold the picture in your head: a revolutionary, a libertarian, and an egalitarian philosopher, all proposing the same cheque. The libertarian wants it to shrink the state; the egalitarian wants it to tame the market; the revolutionary wanted it to repay a theft. They want one instrument for three incompatible ends. That is why a basic income can win a focus group and never a parliament: the coalition that assembles to praise it dissolves the instant anyone has to specify how big it is, how it is paid for, and what it replaces.

The verdict from the trials, part one: it does not make people lazy

For all the philosophy, the single empirical question that has decided basic income's political fate, over and over, is brutally simple: if you give people money for nothing, do they stop working? The fear is ancient and intuitive. The evidence, accumulated now across dozens of studies and several decades, is remarkably consistent — and it says no.

The most rigorous test is also the most recent. Sam Altman's OpenResearch ran an unconditional-cash study in Illinois and Texas from 2020, giving a thousand low-income people a thousand dollars a month for three years, against a control group of two thousand who received fifty — roughly three thousand people enrolled in all. Participants started with average household incomes around twenty-nine thousand dollars, so the payment was a large raise. When the results were published in 2024, the labour-supply effect was real but small: recipients worked about 1.3 to 1.4 hours less per week — the equivalent, across a year, of roughly eight extra days off, less than the gap between American and European holiday norms. The lead economist, Eva Vivalt, called it a "moderate decrease." Crucially, the decrease was not idleness. It was concentrated among single parents, who cut paid hours to do unpaid caregiving, and among the under-thirties, who used the cushion to go back to school — enrolment in education rose thirty-four per cent among the poorest recipients. People with no children and people over thirty barely changed their hours at all.

The world's largest and longest basic-income experiment tells the same story from the opposite side of the planet. The charity GiveDirectly, working with the economists Abhijit Banerjee, Paul Niehaus and Tavneet Suri, has been paying residents of around two hundred Kenyan villages since 2018 — a study designed to run, for one group, a full twelve years. Against the persistent fear that cash breeds dependency, the researchers found "no evidence of UBI promoting laziness." Recipients did not work less; they shifted what they worked on, moving hours out of wage labour and into running their own small businesses. The money did not, as moralists predicted, vanish into drink — if anything, recipients reported seeing fewer of their neighbours drinking daily.

Germany's recent contribution is perhaps the cleanest read of all, because it ran in a wealthy country with a strong existing safety net. From 2021 to 2024, the non-profit Mein Grundeinkommen paid 122 people a comfortable €1,200 a month for three years, with the evaluation handled by the German Institute for Economic Research and universities in Vienna, Frankfurt and Oxford. The headline, in the institute's own words, was that the study "debunks the myth of the social hammock." Recipients did not withdraw from work; they kept putting in around forty hours a week on average. What changed was that more of them felt able to retrain or switch careers, and a third of the money was simply saved. This is, very likely, the kind of programme a participant would describe after living on it for a year or eighteen months — and the labour data is unambiguous: the hammock is a fantasy.

Even Stockton, California, where a guaranteed-income experiment gave 125 residents $500 a month from 2019, found not less work but more. The share of recipients in full-time employment rose from twenty-eight per cent to forty over a single year, while the control group rose only five points. The mechanism is the one the laziness objection cannot see: a person juggling unstable shifts and constant financial emergencies cannot take the unpaid morning off to interview for a better job, cannot fix the car that gets them to work, cannot stop firefighting long enough to climb. A predictable floor bought them the stability to do exactly that. The cash did not replace work; it enabled it.

Exhibit — Given $500 a month, full-time work rose — it did not fall. The control group rose just 5 points — a predictable floor bought people the stability to chase better work, not quit it. Source: SEED, Stockton (2019–21); NPR / Penn CGIR. Compiled by Dan's AI Intel.

The verdict, part two: it reliably makes people healthier and less poor

If the first finding is that cash does little harm to work, the second is that it does considerable good to almost everything else — and here the most powerful evidence is the oldest.

In 1974, the Canadian town of Dauphin, Manitoba, became the site of an experiment unlike any of the modern pilots. Under a programme called Mincome, every household in the town that fell below an income threshold was made eligible for a top-up — not a scattered sample of individuals, but a whole community at once. For decades the data sat in boxes, until the economist Evelyn Forget dug it out and published her findings under the title "The Town with No Poverty." What she found was striking: over the years of the programme, hospitalisations in Dauphin fell by about 8.5 per cent relative to comparable towns, with the largest drops in admissions for accidents, injuries and mental-health diagnoses. More teenagers stayed in school through grade twelve. And the only two groups who measurably reduced their work were new mothers, who took something like maternity leave a generation before it was standard, and adolescents, who were no longer pulled out of school to support the family. Security, it turned out, was a health intervention. People who are not terrified about money get sick less, fight less, and finish school more.

The pattern repeats wherever it is measured. Finland ran a national experiment in 2017 and 2018, paying two thousand unemployed people €560 a month with no strings attached. On the metric the government most wanted to move — getting people into jobs — the effect was negligible, an extra six days of employment over the period. But on wellbeing the effect was large and consistent: recipients reported significantly less stress, better mental health, fewer problems concentrating, and markedly more confidence about their own future and their ability to influence their lives. The verdict that emerged — "happier, but no more likely to find work" — was treated at the time as a disappointment. It is worth sitting with how revealing that framing is. We designed the experiment to test whether free money is a jobs programme, found instead that it is a wellbeing programme, and called the result a failure.

In Kenya, the gains were the most elemental of all. Among GiveDirectly's recipients, the share of households reporting they had gone hungry in the previous month fell sharply — hunger rates dropping from around sixty-eight per cent to fifty-seven. In a setting of genuine deprivation, cash does the obvious thing very well: it buys food, reduces fear, and lets people invest in a future they previously could not afford to think about.

Exhibit — Unconditional cash cut the share of Kenyan households going hungry. In genuine deprivation a cheque does the obvious thing well — it buys food and lifts the fear that comes with hunger. Source: GiveDirectly Kenya UBI study, early results (2023). Compiled by Dan's AI Intel.

Put the two verdicts together and you have a result that ought to settle a great deal. Across forty years and four continents, unconditional cash reliably reduces hardship, improves health, and does not turn anyone into a layabout. By the standard of the thing it was designed to do, basic income works — which makes the retreat of its own champions genuinely puzzling.

The surprises the experiments handed us — including one nobody designed for

A good experiment teaches you what you went looking for. A great one teaches you something you weren't. The basic-income trials have done both, and their accidental findings are where the policy gets genuinely interesting.

The first surprise is that how you give the money may matter as much as whether you do. GiveDirectly's Kenyan study did not just compare cash against no cash; it compared three different ways of delivering the same cash. One group got a long-term monthly payment promised for twelve years. A second got the identical monthly amount, but only for two. A third got a single lump sum of around five hundred dollars, all at once. The textbook expectation was that a steady monthly stream would be gentlest and wisest — that handed a lump sum, poor households would fritter it away. The opposite happened. The lump-sum group started far more businesses and built more lasting wealth than the people who received the same total amount dribbled out monthly, with the lump-sum effect on income reaching something like half of the control group's entire income. A lump you can deploy as capital turns out to do things a trickle never can: it buys the cow, the inventory, the sewing machine. The monthly drip was better at one thing — keeping food on the table and depression down, because a payment that just arrived is a stress that just lifted — but it was the weakest at building a way out.

The second surprise came from the twelve-year arm, and it is subtle but profound. Households that knew the money would keep coming for over a decade behaved differently from those promised only two years — even though, in the first twenty-four months, the two groups received exactly the same cash. The long-horizon group saved and invested far more aggressively. The promise of a durable floor, not just the money itself, changed what people were willing to risk. This is the single most important clue we have about the gap between a pilot and a real policy, and we will return to it, because it cuts both ways.

The third surprise is a disappointment, and it matters precisely because it punctures the optimists' story. Going into the OpenResearch study, the researchers expected that giving people a financial cushion would let them hold out for better jobs — and that the quality of their employment would visibly improve. It didn't. In their own flat words, they found "no effects on the quality of employment whatsoever." Recipients searched more actively and said they cared more about doing meaningful work, but the cushion did not, over three years, translate into better or higher-paid jobs. Cash bought agency and dignity in the moment; it did not buy a rung up the ladder. And the health picture carried its own ambiguity: while younger men in the study sharply reduced dangerous drinking and misuse of painkillers, recipients overall were more likely to be hospitalised and to visit emergency departments — possibly because they could finally afford to seek care they had been deferring, but a reminder that "gave them money, made them healthier" is too simple a sentence for what the data shows.

These are not reasons to dismiss basic income. They are reasons to be honest about what it is. It is a powerful instrument for security, agency and the relief of immediate hardship. It is a weak instrument for mobility, and an untested one for meaning. Which brings us to the half of the ledger no cheque has ever managed to pay.

The half no cheque can pay

Here is the puzzle the data forces on us. The trials are, by their own metrics, a success. So why are the people who funded them backing away — and why would someone who actually lived on a basic income for a year or more tell you, sincerely, that it is not a solution and not something they would wish on a society?

The answer is that the experiments measured the wrong half of what a job is. Decades ago the social psychologist Marie Jahoda, studying the unemployed of the Great Depression, noticed that losing work damaged people in ways that money alone never explained. A wage, she argued, is only the manifest function of employment — the obvious one. Underneath it sit five latent functions that a job quietly supplies and that we barely notice until they are gone. Work imposes a structure on time, the difference between a shaped day and a formless one. It supplies social contact outside the family, a web of colleagues and acquaintances. It binds a person to a collective purpose, the sense of contributing to something larger than oneself. It confers status and identity — the answer to "what do you do," which in most societies is half the answer to "who are you." And it enforces regular activity toward goals set by someone other than yourself, which, Jahoda found, human beings need more than they like to admit; left wholly to our own devices, many of us drift.

The modern evidence has vindicated her almost completely. Meta-analyses pooling decades of studies find that employed people score higher than the unemployed on every one of those five latent functions, and that being deprived of them — not merely of income — is what makes joblessness corrosive to mental health. Now look again at what a basic income does. It restores the manifest function in full: the money is there, unconditionally, reliably. It restores none of the five latent ones automatically. A cheque does not give you colleagues, or a reason the day has a shape, or a place in a shared project, or a status you can name, or a goal that someone is counting on you to meet. It hands you the wage and leaves the architecture of a meaningful life entirely up to you to rebuild from scratch — which a few people manage brilliantly, and many do not.

Exhibit — A cheque restores the wage — and none of the other five things a job supplies. In a post-work economy the binding constraint is meaning, not money — the problem AI actually creates. Source: Jahoda latent-deprivation model; meta-analysis of work's functions (PMC). Compiled by Dan's AI Intel.

This is not a hypothetical worry. It is written in mortality statistics. The economists Anne Case and Angus Deaton gave the phenomenon its name — "deaths of despair" — after documenting a rise in death by suicide, overdose and alcohol among working-class Americans as the stable jobs that once anchored their communities disappeared. The collapse was not only about lost income; transfer payments existed and did not stop it. It was about lost purpose, lost identity, lost membership in a working world. Idleness without a structure to replace it is not leisure; for many people it is a slow harm. This is the thing a basic income, on the current evidence, cannot reach. It is also exactly what someone means when they say a year on a guaranteed income paid the bills and still left them feeling unmoored — grateful for the security and unsettled by the void where the rest of a working life used to be.

That gap is the real reason the AI elite is changing its tune, and it explains Sam Altman's pivot far better than any budget spreadsheet. If artificial intelligence delivers what its builders promise, it will not just reduce incomes that a cheque could top up. It will dissolve the jobs themselves — and with them the time-structure, the colleagues, the purpose, the status, the externally-set goals that those jobs carried as freight. A basic income answers the income problem and walks straight past the meaning problem. And the meaning problem, in a genuinely post-work economy, may be the one that actually decides whether the society holds together.

The bill nobody wants to read

Suppose you set the meaning problem aside and decide a basic income is worth doing for the security and health gains alone. You immediately hit a wall every pilot was designed to avoid showing you: the arithmetic of a real, permanent, universal programme. The economists Hilary Hoynes and Jesse Rothstein laid out the trap with uncomfortable clarity. A basic income large enough to genuinely matter — to lift people clear of poverty — is, in their words, enormously expensive, because by definition you are paying it to everyone, not only the poor. Worse, compared with the targeted programmes a country already runs, a universal grant directs a far larger share of its money to households that are not poor at all: the childless, the non-elderly, the non-disabled, the comfortably middle-income. You are spending vast sums to give money to people who do not need it, in order to preserve the principle that the floor is unconditional. This is Krugman's dilemma stated precisely. Make the grant adequate and it is unaffordable; make it affordable and it is inadequate. There is no version that is both generous and cheap, and the political system has shown no appetite for the taxes the generous version would require.

But the deeper problem with the evidence is not cost — it is that every cheerful pilot result carries an asterisk most reporting omits. The trials are, without exception, dispersed, time-limited, and partial. They give money to a scattered sample of individuals, for a few years, in an economy where almost everyone else is still working and prices are set by normal forces. A real universal basic income would be none of those things. It would be permanent, paid to everyone in a community at once, large enough to live on. And economists have a name for everything that changes when you move from the first world to the second: general-equilibrium effects. When everyone has more money, do prices and rents simply rise to absorb it, eroding the floor you just built? When the income is permanent rather than a three-year windfall, does the modest labour-supply dip seen in the pilots stay modest, or does it deepen as a whole generation grows up knowing the cheque will always be there? When the floor is universal, do social norms about work shift in ways no individual-level study can detect?

We genuinely do not know, and — this is the crucial point — the pilots cannot tell us. They were not built to. Recall the most important finding from Kenya: the group promised twelve years of payments behaved differently from the group promised two, even when the money so far was identical. That single result proves that the duration and credibility of the promise change behaviour — which means a two-year or three-year experiment systematically understates how people would respond to a permanent entitlement. The one real-world case of a permanent, universal, unconditional cash payment is Alaska's Permanent Fund Dividend, paid to every resident from the state's oil wealth since 1982; there, employment did not fall and may even have risen slightly, because the money circulating through the whole community created demand for workers. That is encouraging — but it is one natural experiment, with a dividend far too small to live on, in an economy unlike anyone else's. The honest position is that we have excellent evidence about what a temporary top-up does to a few people and almost no evidence about what a permanent floor would do to everyone. Anyone who tells you the trials have "proven UBI works" is quietly swapping the first claim for the second.

The floor is only as solid as the politics that pours it

There is one more lesson the experiments teach, and it is the one their designers least intended. In 2017 the Canadian province of Ontario launched an ambitious basic-income pilot for four thousand people, a direct descendant of the old Mincome study. A year later a new provincial government, elected on other issues, simply cancelled it — final payments went out in 2019, the data half-gathered, the participants who had reorganised their lives around a three-year commitment left scrambling. The most important thing the Ontario pilot demonstrated had nothing to do with labour supply or wellbeing. It demonstrated that a basic income is only as durable as the next election. A floor you can be voted off is not a floor; it is a ledge.

That fragility connects to the oldest objection of all, the one Rawls aimed at Van Parijs's surfer: reciprocity. Human beings, across cultures, tie their support for sharing to a sense of contribution — a something-for-something ethic in which the community helps those who cannot help themselves and, in return, expects those who can to pitch in. An unconditional income to all, including the able and unwilling, strains that intuition hard. You can argue, as Van Parijs does, that the intuition is unjust — that the surfer is owed his share of a common inheritance no less than anyone. But you cannot legislate as though the intuition does not exist, because it is precisely what a hostile government weaponises when it cancels the programme. The reciprocity objection is not just philosophy; it is the political vulnerability that makes every basic income one election from the axe.

And reciprocity is where the left's own quarrel with basic income lives, which surprises people who assume it is a progressive cause. A strand of the left argues that an unconditional cheque, far from challenging the market, quietly props it up. If the state guarantees everyone a baseline, employers can pay less, confident the public purse will fill the gap — a permanent subsidy to low-wage business, paid for by the taxpayer. And a cash grant, these critics warn, is exactly the kind of thing a small-government movement could use as a Trojan horse: hand everyone a modest cheque and then dismantle public healthcare, public housing, public education on the grounds that people can now "buy their own." Better, this camp argues, to decommodify the essentials — provide health, housing and care directly and universally — and to offer a job guarantee for those who want work, than to hand out money and call the market fixed. "A universal basic income would shore up the market," runs one representative line. "We need ideas that shrink it." Whether or not you agree, it punctures the comfortable assumption that basic income is simply the kind, left-wing option. It is contested from every direction at once.

Where this is heading — and the hinge it turns on

So the champions are walking back, the bill is unpayable in its generous form, the meaning problem is unsolved, and the floor is one election from collapse. Is basic income, then, finished?

Not finished — mutating. The most telling move in the whole debate is the one Sam Altman made: not from supporting basic income to opposing it, but from income to ownership. The new proposals coming out of the AI world are not about redistributing the proceeds of the machines; they are about redistributing the machines themselves — equity stakes for ordinary people, "universal basic compute" that would hand everyone a share of raw AI capability to use or sell, government "wealth funds" that buy a piece of the automation and pay citizens a dividend on it. This is a genuinely more radical idea than a basic income, because it changes who owns the future rather than merely who receives a cheque drawn on it. It is also, conveniently, an idea proposed largely by the people who currently own those machines, which is why it deserves a harder look than its idealistic framing invites — a look we gave it in our episode on why the AI elite quietly gave up on basic income, which traces that pivot and its politics in full. The short version is that "give everyone a slice of the AI" answers the ownership question while saying nothing at all about the meaning one. A dividend is still a cheque. It structures no one's day.

The hinge the whole question turns on is therefore not economic but anthropological. If the AI labs are wrong — if, as the adaptation camp insists, the technology destroys some jobs and creates others, and most people go on trading labour for income much as before — then basic income remains what the evidence already shows it to be: an excellent anti-poverty tool, worth adopting in partial form for the security and health it buys, and no revolution. But if the labs are right, if work really does become economically optional for a large share of humanity, then the binding constraint is not money. We will likely be able to afford to feed everyone; a civilisation that can automate cognition can probably automate abundance. The binding constraint will be meaning — whether a society can manufacture purpose, structure, status and belonging for people the economy no longer needs, when the institution that used to supply all four for free has dissolved. Basic income solves the easy half of that and leaves the hard half untouched. Its own funders sensed this, which is why they are reaching for ownership. But ownership does not solve it either. Nobody, yet, has a serious answer to the meaning problem — and the experiments suggest that the reason is not that we have not spent enough, but that the thing missing was never something money could buy.

Bottom line

After half a century of trials, the verdict on basic income is clear and double-edged. It does the job it was designed for: unconditional cash reliably reduces poverty, improves health, keeps children in school, and — contrary to the most durable objection against it — does not make people stop working. As an anti-poverty instrument, in partial form, it is one of the better-evidenced ideas in social policy. But it does not, and on current evidence cannot, do the job the AI age would actually demand of it. It tops up income; it does not replace the structure, purpose, identity and belonging that work supplies and that a post-labour economy threatens to remove. That is why the very people who funded the biggest experiments are quietly trading "income" for "ownership," and why someone who has lived on a basic income can tell you, without contradiction, that it worked and that it is still not a life. The cheque solves the problem we know how to solve. The future these machines are building would hand us the one we don't.

Sources

Transcript

Sam: The most-studied plan for surviving the AI age works. It genuinely works. It just works on a problem we don't have yet.

Alex: And it fails on the one these machines might create. Cash cures poverty — but it can't buy you a reason to get up in the morning. Which is why its own funders are quietly walking away from it.

Sam: Welcome back to Dan's AI Intel — the show where we try to make sense of the fastest, strangest, most consequential shift any of us is likely to live through. I'm Sam, here with Alex, and today we are getting into universal basic income.

Alex: Which, I know, sounds like a detour into welfare-policy wonkery. Stay with us, because it isn't. The reason this show exists is to help you actually keep pace with the AI revolution — to take the questions that are moving too fast for anyone to track and turn them into something you can hold onto. And this is one of the big ones. Because the moment the people building artificial intelligence started saying "we may have to just pay everyone," a dusty economics debate became a referendum on whether the future they're building is one a human being could actually live in.

Sam: So here's the shape of today. The trigger is simple: the largest cash-handout study in American history just reported its results, and the man who paid for it changed his mind in public the same week. That lets us open a much deeper question than "should welfare be bigger." It's this — if work stops being something most people need to do, what holds a society together? We're going to travel across fifty years and four continents of real experiments, from a Canadian town in the seventies to villages in Kenya running a twelve-year trial right now. We'll meet a revolutionary, a libertarian, and a philosopher who all wanted the exact same cheque for completely opposite reasons.

Alex: And we will get to the turn that genuinely surprised me — the reason a person who actually lived on a guaranteed income for over a year will tell you, sincerely, that it worked and that it is still not a life. There's a half of this that every cheerful headline misses, and once you see it, you can't unsee it.

Sam: One quick thing before we dive in: if you've been enjoying the show, hit follow wherever you're listening. It's free, it's one tap, and it means the next one lands in your feed automatically. Okay. Let's start with why a fight about welfare is secretly a fight about the next ten years.

Alex: So for about two centuries, the idea of just handing everyone money sat at the very edge of economics. A few philosophers argued about it. A handful of towns tried it. Nobody, anywhere, adopted it at full strength. It was a thought experiment with a long beard.

Sam: And then it stopped being a curiosity. What changed?

Alex: The people building artificial intelligence picked it up. And that completely changes the character of the question. When the engineers who expect their own machines to do a huge share of human labour start saying a no-strings income might be the only humane way through — that's not a welfare argument anymore. It's a structural one.

Sam: Right, because it's coming from the inside. It's not bleeding-heart campaigners saying "be more generous." It's the architects of the thing saying "we might break the deal that everyone's life is built on."

Alex: Exactly that. The defining claim of this whole moment — and it's the labs making it, not their critics — is that within years, not generations, software will do cognitive work that today needs a salaried human. And if that's even half true, the twentieth-century arrangement where most people trade their labour for income comes under real strain.

Sam: And universal basic income is the most developed answer anyone's actually put on the table for that.

Alex: It's the most worked-out response we have. So this stops being a side-quest in welfare policy and becomes a direct test — is the future these companies are building one a person could genuinely live in?

Sam: Okay, but I want to push on something straight away, because it's the thing that makes me hopeful here. Normally when we talk about the AI future on this show, we're squinting into fog. We're guessing. This one's different, isn't it?

Alex: This one's different. That's the good news. Unlike almost every other question about the AI economy — where we're genuinely speculating — this one has been studied for fifty years. On four continents. With real money and real randomised control groups. We are not arguing in the dark here.

Sam: So we actually have data.

Alex: We have unusually good data. And here's the hook of the whole thing — the data points somewhere genuinely surprising. Somewhere that should unsettle the people selling basic income and the people sneering at it. Both camps are comfortable, and both camps are about to be made uncomfortable.

Sam: Love that. Okay — so if we've got fifty years of evidence, let's start with the people. Who actually signed up for this idea?

Alex: Start with the roll call, because it's more impressive than the sceptics admit, and more fragile than the believers like to mention. Elon Musk has said for years that advanced AI and robotics will wipe out the need for most jobs, and that some guaranteed income becomes unavoidable as a result. He doesn't even like the word "basic" — he calls it "universal high income," and paints it as the on-ramp to a kind of Star Trek future of post-scarcity abundance.

Sam: That's a very Musk framing. Not "a safety net," but "the gateway to utopia."

Alex: Right. Then Andrew Yang built an entire 2020 presidential campaign on this — a thousand dollars a month, he called it the "Freedom Dividend," and he said flatly that "AI will wind up funding universal income" and the only question was how fast. Mark Zuckerberg stood up at his Harvard commencement address in 2017 and told the graduates the country should "explore ideas like universal basic income."

Sam: So this isn't fringe. These are the most powerful people in the industry.

Alex: And it goes deeper than the CEOs. Geoffrey Hinton — and people should know that name, he's the computer scientist whose work basically underpins the entire deep-learning boom, the man who left Google in 2023 specifically to warn about where this is all heading —

Sam: The "godfather of AI" guy.

Alex: That's him. He went further than most. He's said he was actually consulted by the UK government, and that he told them a basic income was a sensible response to the job losses he expects. And Chris Hughes, one of the co-founders of Facebook, put real money into guaranteed-income research. So when you add it up: at least half a dozen of the most prominent figures in and around the field have, at some point, called a basic income not just a nice idea but inevitable.

Sam: Okay, that is a heavyweight list. So where's the fragility you mentioned? Because you said it was more fragile than the believers admit.

Alex: The fragility is one man. Sam Altman. The head of OpenAI. And he did something rarer than just endorse the idea — he tested it. Through a research outfit called OpenResearch, Altman backed what became the largest unconditional-cash study in modern American history. Put his own money behind actually finding out.

Sam: Which is exactly what you'd want, right? Don't just preach it — fund the experiment.

Alex: It's the honourable move. And then, just as the results landed, he changed his mind in public. His words: "I no longer believe in universal basic income as much as I once did." A fixed monthly cheque, he said, "does not get us what we're really going to need for this next phase."

Sam: Wait. The person who paid for the biggest test of the idea looked at the results and backed off it? That's the tell, isn't it? That's not nothing.

Alex: It's the first crack in the consensus, and it's a loud one. And his attention didn't just retreat — it moved. From income to ownership. Instead of a cheque, he started talking about equity stakes in the machines. Something he calls "universal basic compute" — giving every person a slice of AI's raw capability. Even a half-serious idea to measure the world's total AI output in tokens and hand a share of it out to all eight billion humans.

Sam: Hold on, that's a real shift in the underlying logic. A cheque is "here's money the machines made." Ownership is "here's a piece of the machine itself."

Alex: You've put your finger on the whole rest of the episode, honestly. Hold that distinction — income versus ownership — because it comes back at the end and it matters enormously. But Altman isn't the only crack. Paul Krugman, the Nobel laureate, has been blunt: "I'm not a UBI guy." His reasoning is sharp — a basic income generous enough to actually matter would cost more than seems politically possible for decades, and a version cheap enough to pass would be too small to live on.

Sam: So either it's unaffordable or it's useless. That's a tight little trap.

Alex: Hold onto that trap too — it has a name and we'll come back to it. The International Labour Organization published a paper arguing, in plain terms, that a basic income is "not the answer" to AI's impact on work. And then there's a whole school of thought — call it the adaptation camp — that points out every previous technological revolution destroyed particular jobs while creating more in total. And that planning to just pay everyone off assumes a kind of helplessness in ordinary workers that history has never actually borne out.

Sam: So let me say back what the honest tally is. The list of believers is real, and it's seriously smart people. But their conviction is fracturing exactly as the actual evidence shows up. And the guy who funded the experiment is running away the fastest.

Alex: That's the picture. And that is a very strange thing for a winning idea to do. Winners don't usually get abandoned by their own backers the moment the results come in. To understand why, we have to do something boring-sounding that turns out to be the key to the entire argument.

Sam: Uh oh. What's the boring thing?

Alex: We have to be precise about what "basic income" even means. Because — and this is genuinely half the reason the public fight is so confused — two camps are using one phrase for two completely different machines.

Sam: Okay, so when you say two different machines — break that down for me. What is a universal basic income, in the textbook sense?

Alex: In its pure, textbook form it has four properties, all at once. It's universal — everyone gets it, rich and poor alike. It's unconditional — no work requirement, no means test, nothing you have to prove or behave a certain way to qualify. It's individual — paid to people, not households. And it's periodic — a regular cash payment, not a one-off grant, and not a voucher you can only spend on approved things.

Sam: So all four together. And if you drop one?

Alex: If you strip away any one of those four, you've got something else. A targeted benefit. A tax credit. A one-time stake. The purity is the whole point — the promise is a floor below which no one can ever fall, that nothing you do can take away from you.

Sam: Got it. Four properties, non-negotiable. So where's the slippery part?

Alex: The slippery part is the word "basic." Because "basic" is secretly a claim about scale. It means an income at the level you'd need to cover basic needs — roughly subsistence, somewhere around a national poverty line. And here's the thing that reframed this entire topic for me: by that strict standard, almost none of the experiments people point to as "UBI trials" tested a basic income at all.

Sam: Wait, what? None of them? But people cite these studies constantly.

Alex: They tested partial basic incomes. Think of it like this — there's a difference between a top-up and a replacement. Five hundred dollars a month in Stockton. A thousand in the OpenResearch study. Twelve hundred euros in Germany. Those are sums that supplement a life. They sit on top of whatever else you've got going. They don't replace a wage.

Sam: So they're a cushion, not a foundation.

Alex: That's exactly the right image. A cushion, not a foundation. A genuinely full basic income in the United States — one a person could actually live on without working at all — would be something on the order of twelve to fifteen thousand dollars per adult, per year. And a lot more for a family.

Sam: So the gap between what got tested and what's actually being promised is enormous.

Alex: It's most of the reason the evidence is easier to read than the politics. We've got strong data on what a top-up does. We have almost none on what a true, permanent, replace-your-wage floor would do. Those are two different questions, and people swap them constantly.

Sam: Okay, and I feel like there are a bunch of similar-sounding terms that get jumbled in here too. Negative income tax, guaranteed income, job guarantee. Are those all the same thing or different things?

Alex: Different machines again, and it's worth thirty seconds to keep them straight, because politicians blur them on purpose. A negative income tax — that's the version Milton Friedman proposed back in 1962 — delivers the floor through the tax system. So instead of paying tax, people below a certain threshold receive a payment that's a fraction of how far short they fell.

Sam: So the tax office pays you, rather than you paying it.

Alex: Right, it runs in reverse below the line. Then a guaranteed or minimum income — that's the means-tested welfare model most countries already run. It's a floor, but it claws back as you earn. Sometimes at close to a dollar lost for every dollar you make.

Sam: Oh, that's the thing that traps people, isn't it. Where taking a job barely pays because they yank the benefit away at the same speed.

Alex: That's the notorious "welfare trap," exactly. You earn a dollar, you lose nearly a dollar of support, so why bother. And then a job guarantee — which is favoured more on the left — is the complete opposite philosophy. It's not cash for nothing. It's a standing offer of paid public work to anyone who wants it, with the government as the employer of last resort.

Sam: So that's "we'll always give you a job," versus "we'll always give you money." Genuinely opposite instincts.

Alex: Opposite instincts. And here's the subtlety that trips up almost every conversation. Friedman himself pointed out something beautiful — that a universal basic income and a negative income tax are, mathematically, almost the same thing.

Sam: How can they be the same? One pays everyone, one only pays the poor.

Alex: Watch. You pay a basic income to everyone, and then you tax it back from the people who don't need it. The final distribution — who ends up with what — is identical to a negative income tax that only ever pays the people below the line. Same money ends up in the same pockets.

Sam: Huh. So on a spreadsheet they're twins.

Alex: On a spreadsheet, twins. And the economist Philippe Van Parijs — he's the idea's most rigorous modern defender, we'll hear a lot more from him in a second — he agreed they're economically equivalent. But he insisted the timing changes behaviour.

Sam: The timing? Same money, but when you get it matters?

Alex: It matters enormously, and this is one of the most quietly profound things in the whole subject. A basic income arrives before you act. Every month. Unconditionally. You hold the floor in your hand and you plan around it. A negative income tax arrives after — at tax-filing time, as a reconciliation, a settling-up.

Sam: So one is security you can feel, and the other is a refund you eventually get.

Alex: You just wrote the line better than I was going to. The first gives you security you can feel and use. The second gives you the exact same money with none of the psychological permission. And that distinction — security held in advance, versus compensation paid in arrears — turns out to drive a huge amount of what recipients actually do with their lives.

Sam: That's such a good point, because it means the same dollar can change your behaviour or not, depending entirely on whether you trust it's coming.

Alex: And it explains why two people can argue about "basic income" and be describing completely different things. One's defending an unconditional cheque to everyone. The other's attacking a means-tested handout to the poor. And they never notice they're describing different machines. Keep the four properties in mind, keep "partial versus full" in mind, and honestly most of the noise just resolves into one real, narrow disagreement.

Sam: Okay, that's the vocabulary sorted. But you teased something about a revolutionary, a libertarian, and a philosopher all wanting the same cheque. I need that story.

Alex: It's the strangest fact about this whole idea. Its most committed champions have spent two hundred years agreeing on the policy while disagreeing about literally everything else. This is not a left idea. It is not a right idea. It's an idea that the left and the right keep arriving at from opposite directions, for opposite reasons.

Sam: Which sounds like a superpower. If everyone wants it —

Alex: It sounds like a superpower. Hold that thought, because it's actually the fatal flaw in disguise, and that's the payoff of this whole section. But first the three characters. The first recognisable version came from Thomas Paine — the pamphleteer of the American and French revolutions — in a 1797 essay called Agrarian Justice.

Sam: 1797. So this is properly old.

Alex: Properly old. And Paine's argument wasn't about charity, and it wasn't about efficiency. It was about property and theft. His reasoning: in the state of nature, the earth was the common inheritance of everyone. Then the moment land became private property, the people left without any were made worse off than they would have been in the wild — robbed of a birthright they never agreed to give up.

Sam: Oh, that's a sharp framing. So the landowners don't owe the landless a kindness —

Alex: They owe them a debt. That's exactly Paine's move. Not charity, a debt. And he proposed paying it as a lump sum to every person when they turned twenty-one — fifteen pounds, something like two and a half thousand of today's pounds — plus an annual pension from the age of fifty. So that's the very first basic income, and it's justified entirely as a right. Compensation for an enclosure none of us consented to.

Sam: Okay, revolutionary, check. Who's the libertarian?

Alex: This is the fun one, because it's the least likely patron imaginable. Milton Friedman. The patron saint of free markets. And he proposed his negative income tax not despite his conservatism but because of it.

Sam: How does the free-market guy end up wanting to hand out money? That feels backwards.

Alex: Here's how it tracks. To Friedman, the sprawling apparatus of the welfare state — all its dozens of programmes, its armies of caseworkers, its rules about what poor people could buy and how they had to behave — that whole machine was an offence. Against efficiency and against freedom. So his pitch: replace the entire thing with a single cash transfer. You'd help people more, it'd cost less, and you'd stop treating adults like wards of the state.

Sam: So for him the cash isn't generosity. It's the least bossy thing a government can do.

Alex: That's the right-wing case in one sentence — give people the cash and let them, not a bureaucrat, decide what they need. It's the least paternalistic option on the table. And then the third character, in 1995, the Belgian philosopher Philippe Van Parijs gives the idea its most ambitious modern grounding, in a book called Real Freedom for All.

Sam: And what's his angle? We've had a right and a debt. What's left?

Alex: He redefines freedom itself. His argument: formal freedom — the legal right to do something — is worthless if you don't have the means to actually do it. Picture someone who's free to start a business but has no savings. Free to walk out of an abusive job but has rent due Friday. Free to write a novel but can't afford to eat. Are they really free?

Sam: No. They've got the permission slip and none of the power.

Alex: That's it precisely. So real freedom, Van Parijs says, is the effective capacity to actually live as you choose. And the unconditional grant is the thing that delivers it to everyone, not just the people who already own property. And he cheerfully accepted the most provocative version of this — that even a surfer who chooses to ride waves at Malibu instead of working would be owed his share.

Sam: Ohh. Okay, that's the line that gets people's blood up, isn't it. The surfer who just won't work, and you still have to pay him.

Alex: That is the exact line that drew the fire. The philosopher John Rawls, among others, objected — you can't ask the diligent to subsidise the voluntarily idle without breaking some basic principle of reciprocity. Of fairness. And we are going to come back to that surfer, because the reciprocity objection turns out to be the hinge of the entire debate. Tuck him away in your mind.

Sam: So let me hold the three of them together. A revolutionary, a libertarian, and an egalitarian philosopher, all proposing the same cheque.

Alex: All proposing the same cheque — and this is the payoff. The libertarian wants it to shrink the state. The egalitarian wants it to tame the market. And the revolutionary wanted it to repay a theft. They want one instrument to serve three completely incompatible ends.

Sam: And that's why it can win a focus group but never a parliament.

Alex: You just nailed the whole thing. The coalition that assembles to praise basic income dissolves the instant anyone has to actually specify how big it is, how you pay for it, and what it replaces. Everyone loves it in the abstract. Nobody can agree on it in the concrete. So the breadth of support that looks like its great strength —

Sam: — is exactly why it never gets built. It's loved to death by people who'd never tolerate each other's version of it.

Alex: That's the trap of universal appeal. Okay. We've done the philosophy, we've done the cast of characters. But this show is supposed to be evidence-led, and we said up front the whole point here is that we actually have fifty years of real data. So let's go to the trials. And let's start with the single question that has decided this idea's political fate over and over again. For all the philosophy, the single empirical question that has decided basic income's fate, over and over, is brutally simple. If you give people money for nothing, do they stop working?

Sam: And that's the ancient fear, right? It's the gut reaction everyone has. Free money equals hammock. Surely some people just down tools.

Alex: It's ancient and it's intuitive — and the evidence, now piled up across dozens of studies over several decades, is remarkably consistent. And it says no.

Sam: Flatly no?

Alex: Let me give you the most rigorous test, which is also the most recent — the OpenResearch study, Altman's one. They ran an unconditional-cash study in Illinois and Texas, starting in 2020. They gave a thousand low-income people a thousand dollars a month, for three years. Against a control group of two thousand who got fifty dollars a month. So about three thousand people enrolled in total.

Sam: And these are genuinely low-income people, so a thousand a month is a serious amount to them.

Alex: It's a big raise. They started with average household incomes around twenty-nine thousand dollars. So a thousand a month is real money. And when the results published in 2024, the effect on how much people worked was real — but small. Recipients worked about 1.3 to 1.4 hours less per week.

Sam: 1.3 hours. That's... what, like leaving twenty minutes early some days? That's barely anything.

Alex: Put it on an annual scale and it's even clearer — across a whole year it's roughly eight extra days off. Which is less than the gap between American and European holiday norms. The lead economist, Eva Vivalt, called it a "moderate decrease."

Sam: So nobody quit. They just... eased off by a sliver.

Alex: And here's the part that matters more than the number. The decrease wasn't idleness. It was concentrated in two specific groups. Single parents — who cut paid hours to do unpaid caregiving, to look after their own kids. And the under-thirties — who used the cushion to go back to school. Enrolment in education rose thirty-four per cent among the poorest recipients.

Sam: Oh, so the "working less" is actually "parenting" and "studying." That's a completely different story than the lazy one.

Alex: Completely different. And the people with no kids, and the people over thirty? They barely changed their hours at all. So the headline "they worked less" is technically true and deeply misleading.

Sam: That's a really important distinction. Okay, that's America. You keep saying four continents — give me another one, somewhere totally different, because one study could be a fluke.

Alex: Fair, and you should always want the replication. So go to the opposite side of the planet. The world's largest and longest basic-income experiment is being run right now by the charity GiveDirectly, working with the economists Abhijit Banerjee, Paul Niehaus and Tavneet Suri. They've been paying residents of around two hundred Kenyan villages since 2018. And one group's version is designed to run a full twelve years.

Sam: Twelve years. That's not a pilot, that's a chunk of a life.

Alex: It's the closest thing we've got to the real deal, which becomes important later. And against the persistent fear that cash breeds dependency, the researchers found — their phrase — "no evidence of UBI promoting laziness." People didn't work less. They shifted what they worked on. They moved hours out of wage labour and into running their own small businesses.

Sam: So entrepreneurship, not idleness. And what about the other classic moral panic — that they'd just drink it?

Alex: The "they'll spend it on booze" fear. It didn't happen. If anything, recipients reported seeing fewer of their neighbours drinking daily. The money did not vanish into drink.

Sam: I love that the data just quietly demolishes the thing everyone's so sure about. Okay, third one. Hit me with another.

Alex: This one might be the cleanest read of all, because it ran in a wealthy country with a strong existing safety net — so you can't say "well, of course poor people use it well." Germany. From 2021 to 2024, a non-profit called Mein Grundeinkommen paid 122 people a comfortable twelve hundred euros a month, for three years. And the evaluation was done seriously — the German Institute for Economic Research, plus universities in Vienna, Frankfurt and Oxford.

Sam: And twelve hundred euros in Germany is a genuinely comfortable top-up.

Alex: It's comfortable. And the institute's own headline phrase was beautiful. They said the study "debunks the myth of the social hammock."

Sam: The social hammock. That's the exact image, isn't it — people lounging around on free money.

Alex: And what they actually found: recipients didn't withdraw from work. They kept putting in around forty hours a week, on average. What changed was that more of them felt able to retrain, or switch careers. And a third of the money was simply saved. Honestly, this is very likely the kind of programme a person would describe after living on it for a year or eighteen months. And the labour data is just unambiguous — the hammock is a fantasy.

Sam: Okay so America, Kenya, Germany — all saying the same thing. Is there one where they actually worked more? Because that would really put a stake through it.

Alex: There is, and it's the kicker. Stockton, California. A guaranteed-income experiment gave 125 residents five hundred dollars a month from 2019. And they found not less work but more. The share of recipients in full-time employment rose from twenty-eight per cent to forty in a single year — while the control group rose only five points.

Sam: That's the opposite of the prediction. They gave people money and more of them got full-time jobs. How does that even work?

Alex: This is the mechanism the laziness objection literally cannot see, and it's the most important idea in this section, so let me slow down on it. Picture someone juggling unstable shifts and constant financial emergencies. That person cannot take an unpaid morning off to go interview for a better job. They can't afford to fix the car that gets them to work. They can't stop firefighting long enough to ever climb.

Sam: Right — because every spare hour is spent plugging the next leak.

Alex: Every hour. Now give them a predictable floor. Suddenly they can take that morning. They can fix the car. They can stop firefighting long enough to actually reach for the better rung. The cash didn't replace work.

Sam: It enabled it. The money bought them the stability to go and earn more. That's genuinely counterintuitive — I'd have bet the other way.

Alex: Most people would. And that's the first verdict, rock solid across four continents and forty years: free money does not make people lazy. If anything, for the people right at the edge, it frees them to work better. But that's only half of what these trials found. The other half is, if anything, even more powerful — and it's the oldest evidence of all.

Sam: Older than the seventies? You keep going further back.

Alex: 1974. A Canadian town called Dauphin, in Manitoba. And this experiment was unlike any of the modern pilots in one crucial way. Under a programme called Mincome, every household in the town that fell below an income threshold was made eligible for a top-up. Not a scattered sample of individuals dotted around — a whole community, all at once.

Sam: Oh, that's a different thing entirely. That's the whole town living in it together.

Alex: Which matters more than it sounds, and we'll see why later. For decades the data just sat in boxes, unanalysed. Until an economist named Evelyn Forget dug it out and published her findings. And she gave it a title I love — "The Town with No Poverty."

Sam: Great title. What did she find in the boxes?

Alex: Something striking. Over the years of the programme, hospitalisations in Dauphin fell by about 8.5 per cent, relative to comparable towns. And the biggest drops were in admissions for accidents, injuries, and mental-health diagnoses.

Sam: Wait, hospital visits dropped? From giving people money? That's not the kind of thing I'd expect cash to touch at all.

Alex: That's the surprise. Think of it like this — being broke is a chronic stress, and chronic stress is a physical illness with a delay. Take the financial terror away, and the accidents, the breakdowns, the stress injuries all tick down. Security, it turns out, is itself a health intervention. And more teenagers stayed in school through grade twelve.

Sam: And the working-less thing?

Alex: Same pattern as everywhere. The only two groups who measurably reduced their work were new mothers — who took something like maternity leave a generation before it was standard — and adolescents, who were no longer being pulled out of school to support the family. So even the "worked less" is, again, mothers and teenagers. People who are not terrified about money get sick less, fight less, and finish school more.

Sam: That reframes the whole thing. We keep measuring whether it's a jobs programme, and it keeps turning out to be a health programme.

Alex: Hold exactly that thought, because Finland makes it explicit. Finland ran a national experiment in 2017 and 2018 — paid two thousand unemployed people five hundred and sixty euros a month, no strings. And on the metric the government most wanted to move — getting people into jobs — the effect was negligible. An extra six days of employment over the whole period.

Sam: Six days. So by the government's own scorecard, it flopped.

Alex: By that scorecard, a flop. But on wellbeing, the effect was large and consistent. Recipients reported significantly less stress, better mental health, fewer problems concentrating, and markedly more confidence about their own future and their ability to influence their own lives. The verdict that went around at the time was "happier, but no more likely to find work" — and it was treated as a disappointment.

Sam: But sit with that for a second. That's bananas. We designed the test to see if free money is a jobs programme. We found out it's a wellbeing programme. And we filed that under "failure."

Alex: That's the line I keep coming back to. It's worth really sitting with how revealing that framing is. We asked the wrong question, got a profound answer to a different one, and called it a let-down because it didn't answer the question we asked.

Sam: That's almost a parable about the whole debate.

Alex: It kind of is. And in Kenya, the gains were the most elemental of all — just the basics of survival. Among GiveDirectly's recipients, the share of households reporting they'd gone hungry in the previous month fell sharply. Hunger dropped from around sixty-eight per cent to fifty-seven.

Sam: So in a place of genuine deprivation, the money does the obvious thing extremely well. It buys food.

Alex: It buys food, it reduces fear, and it lets people invest in a future they previously could not afford to even think about. So put the two verdicts together and you've got a result that ought to settle a great deal. Across forty years and four continents, unconditional cash reliably reduces hardship, improves health, and turns nobody into a layabout. By the standard of the thing it was designed to do —

Sam: — it works. It just works. Which makes the retreat of its own champions even weirder than when we started. If the evidence is this good, why is Altman running?

Alex: That is precisely the cliff we're now standing on. But before we get to why they're running, the experiments handed us a few surprises nobody designed for. And one of them is going to change how you think about giving anyone money ever again. A good experiment teaches you what you went looking for. A great one teaches you something you weren't even asking. And the basic-income trials have done both — but it's the accidental findings where this gets genuinely fascinating.

Sam: Okay, accidental findings are my favourite kind. What's the first one?

Alex: The first surprise is that how you give the money may matter as much as whether you give it at all. And this comes back to Kenya, the GiveDirectly study, because they did something clever. They didn't just compare cash against no cash. They compared three different ways of delivering the exact same cash.

Sam: Same total amount, three different shapes?

Alex: Three shapes. Group one got a long-term monthly payment, promised for twelve years. Group two got the identical monthly amount, but only for two years. And group three got a single lump sum — about five hundred dollars, all at once.

Sam: Okay, and what would you predict? Because my instinct says the steady monthly drip is the responsible one. Give a poor household a big lump and surely they blow it.

Alex: That is the textbook expectation, and it's exactly what most people assume — that a steady stream is gentlest and wisest, and that handed a lump sum, poor households fritter it away. And the opposite happened.

Sam: The opposite. So the lump sum won?

Alex: The lump-sum group started far more businesses and built more lasting wealth than the people who got the same total money dribbled out monthly. The lump-sum effect on income reached something like half of the control group's entire income. Here's the intuition — a lump you can deploy as capital does things a trickle never can. It buys the cow. The inventory. The sewing machine. You can't buy a cow twenty dollars at a time.

Sam: Right, because the trickle gets eaten by daily life before it can ever become a tool. You need the whole lump in one hand to actually buy the thing that changes your income.

Alex: That's the mechanism exactly. And the magnitude here is genuinely startling — the lump-sum effect on income reached something like half of the control group's entire income. Sit with that. The single five-hundred-dollar payment, deployed as capital, generated a return so large it was equivalent to about half of what a control household earned in total. That's not a subsidy. That's a launchpad.

Sam: So the same money, in a different shape, stops being "help getting by" and becomes "the seed of a business." That's a wild difference from literally just changing the timing.

Alex: It's the difference between feeding someone and capitalising them. Though — and this is the honest balance — the monthly drip was better at one thing. Keeping food on the table and depression down. Because a payment that just arrived is a stress that just lifted, every single month. But it was the weakest at building an actual way out.

Sam: So there's a real trade-off baked in. The steady stream soothes; the lump transforms. That's a genuinely useful thing to know if you're ever designing one of these.

Alex: It rewrites the design question. And the second surprise comes from that twelve-year arm, and it's subtle but I'd say profound. Households that knew the money would keep coming for over a decade behaved differently from the ones promised only two years. Even though — and this is the key — in the first twenty-four months, both groups got exactly the same cash.

Sam: Hang on. Same money, same first two years. The only difference is one group knows it continues and the other doesn't. And they behave differently?

Alex: The long-horizon group saved and invested far more aggressively. So it wasn't the money itself that changed their behaviour. It was the promise. The belief that the floor was durable, that it would still be there in ten years, changed what people were willing to risk today.

Sam: That is profound. Because it means the thing doing the work isn't the cash — it's the confidence. The trust that it's permanent.

Alex: And file that carefully, because it's the single most important clue we have about the gap between a pilot and a real policy. It's going to come back and it cuts in a direction the optimists won't love. But let me give you the third surprise, because it's a disappointment, and it matters precisely because it punctures the optimists' story.

Sam: Okay, brace me. What did they hope for that didn't happen?

Alex: Going into the OpenResearch study, the researchers expected that giving people a financial cushion would let them hold out for better jobs. That the quality of their employment would visibly improve — they'd refuse the bad gig and wait for the good one.

Sam: That makes total sense. With a cushion, you don't have to take the first lousy job that comes along.

Alex: It makes total sense, and it didn't happen. Their own flat words: they found "no effects on the quality of employment whatsoever."

Sam: Whatsoever. Oof.

Alex: Recipients searched more actively. They said they cared more about doing meaningful work. But the cushion did not, over three years, translate into better or higher-paid jobs. Cash bought agency and dignity in the moment. It did not buy a rung up the ladder.

Sam: So that's a real limit. It's a powerful thing for right now, today, this month — and a weak thing for "where am I in five years."

Alex: That's the precise shape of it. And even the health picture carried an ambiguity I want to be honest about. While younger men in the study sharply reduced dangerous drinking and misuse of painkillers — genuinely good — recipients overall were actually more likely to be hospitalised, and to visit emergency departments.

Sam: Wait, more hospital visits? That cuts against the Dauphin finding.

Alex: It does, and the likely explanation is almost hopeful — they could finally afford to go and get care they'd been putting off. They'd been deferring the doctor because they couldn't pay, and now they went. But it's a reminder that "gave them money, made them healthier" is too simple a sentence for what the data actually shows.

Sam: Okay so let me try to land this section honestly, because it's not all one direction. These aren't reasons to dismiss basic income. They're reasons to be precise about what it actually is.

Alex: That's exactly the right way to hold it, and you've said it better than the summary I had. It's a powerful instrument for security, for agency, for relieving immediate hardship. It's a weak instrument for mobility — for climbing. And it's an untested instrument for meaning. Which brings us to the half of the ledger that no cheque has ever managed to pay. And honestly, this is the part of the episode I find hardest to shake. Here's the puzzle the data forces on us. The trials are, by their own metrics, a success. So why are the people who funded them backing away? And — this is the human version of the same question — why would someone who actually lived on a basic income for a year or more tell you, sincerely, that it's not a solution? That it's not something they'd wish on a society?

Sam: Yeah, that's the thing I genuinely can't square. If I got a comfortable cheque every month, no strings, I think I'd be thrilled. Where does the "and yet it's not a life" come from?

Alex: The answer is that the experiments measured the wrong half of what a job actually is. And to see it, you have to go back to a social psychologist named Marie Jahoda. She was studying the unemployed during the Great Depression, and she noticed something. Losing work damaged people in ways that money alone never explained. People with savings, people whose bills were covered — they still came apart.

Sam: So it wasn't just about being broke. Something else was breaking.

Alex: Something else. And she made a distinction I think about constantly. A wage, she said, is only the manifest function of employment. The obvious one. The one on the surface. But underneath it sit five latent functions — five things a job quietly supplies that we barely even notice until they're gone.

Sam: Okay, hidden functions of a job. Walk me through the five.

Alex: One — a job imposes a structure on time. The difference between a day that has a shape and a day that's a formless blob. Two — it supplies social contact outside your family. A whole web of colleagues and acquaintances you'd never otherwise have. Three — it binds you to a collective purpose. The sense that you're contributing to something bigger than just yourself. Four — it confers status and identity. It's the answer to "so what do you do?" — which in most societies is half the answer to "who are you?" And five — it enforces regular activity toward goals that someone other than you has set.

Sam: That last one's interesting. Goals set by someone else. Why does that matter? Surely we'd all rather set our own.

Alex: We'd like to think so. But Jahoda found that human beings need that external structure more than they like to admit. Left wholly to our own devices, a lot of us just... drift. The deadline someone else imposed is secretly holding your week together.

Sam: Okay, that one stings a little, because it's true. So those are her five — structure, contact, purpose, status, and external goals. And has that actually held up? Because Depression-era psychology could just be of its time.

Alex: It's held up almost completely. Modern meta-analyses — pooling decades of studies — find that employed people score higher than the unemployed on every single one of those five. And crucially, that being deprived of them, not merely of income, is what makes joblessness so corrosive to mental health.

Sam: So now do the move. Hold a basic income up against those five.

Alex: This is the whole episode in one beat. A basic income restores the manifest function in full. The money's there — unconditionally, reliably. Perfect. And it restores none of the five latent ones. Automatically? Zero. A cheque does not give you colleagues. It doesn't give you a reason the day has a shape. It doesn't hand you a place in a shared project, or a status you can name, or a goal that somebody's counting on you to hit.

Sam: It pays the wage and just... leaves the rest of it blank. It hands you the money and says "the architecture of a meaningful life? You build that yourself, from scratch."

Alex: That's the devastating version of it, and it's exactly right. Which a few people manage brilliantly. And many people do not. And this isn't a hypothetical worry — this is the part that genuinely got under my skin — it's written in mortality statistics.

Sam: Mortality statistics. You mean people are dying from this gap?

Alex: The economists Anne Case and Angus Deaton gave the phenomenon its name. "Deaths of despair." They documented a rise in death — by suicide, by overdose, by alcohol — among working-class Americans, as the stable jobs that once anchored their communities disappeared.

Sam: And the crucial thing, I'm guessing, is that this wasn't only about the lost paycheque.

Alex: That's the whole point, and it's the bit that should stop the optimists cold. The collapse was not only about lost income. Transfer payments existed. Welfare existed. And they did not stop it. It was about lost purpose. Lost identity. Lost membership in a working world. Idleness without a structure to replace it isn't leisure. For a lot of people, it's a slow harm.

Sam: So the cheque was there, in some form, and people died anyway — because the cheque couldn't touch the thing that was actually killing them.

Alex: That is exactly the thing a basic income, on the current evidence, cannot reach. And it's exactly what someone means when they tell you a year on a guaranteed income paid the bills and still left them feeling unmoored. Grateful for the security. Unsettled by the void where the rest of a working life used to be.

Sam: That's the friend-who-lived-it version, isn't it. Not "it didn't work." But "it worked, and I was still adrift."

Alex: Both true at once. And here's why this is the real reason the AI elite is changing its tune — it explains Altman's pivot far better than any budget spreadsheet does. If artificial intelligence delivers what its builders are promising, it won't just reduce incomes that a cheque could top up. It will dissolve the jobs themselves. And with them, all five of those things — the time-structure, the colleagues, the purpose, the status, the goals — that the jobs were carrying as freight the whole time.

Sam: So the machines don't just take the wage. They take the scaffolding the wage was bolted to. And basic income only ever replaces the wage.

Alex: A basic income answers the income problem and walks straight past the meaning problem. And the meaning problem, in a genuinely post-work economy, may be the one that actually decides whether the society holds together at all. That — for my money — is the deepest finding in this entire subject. Not the cost. Not the labour-supply numbers. This.

Sam: Okay. I need a second after that one, honestly. Because it completely flips the frame. We started thinking the hard question was "can we afford it." And the hard question turns out to be "can we survive it." So — let's actually do the affordability question, because I still want to know. Even if meaning is the deeper problem, is the cheque itself even payable?

Alex: Suppose you set the meaning problem aside — park it — and decide a basic income is worth doing for the security and the health gains alone. Which, given what we just heard, is a perfectly reasonable call. You immediately hit a wall that every single pilot was carefully designed to never show you. The arithmetic of a real, permanent, universal programme.

Sam: The pilots were designed to hide the cost?

Alex: Not hide it maliciously — but a pilot pays a few thousand people, so it never has to confront the actual bill. And the economists Hilary Hoynes and Jesse Rothstein laid out the trap with uncomfortable clarity. A basic income large enough to genuinely matter — to lift people clear of poverty — is, in their words, enormously expensive. Because by definition you're paying it to everyone. Not just the poor.

Sam: Right, that's the universal part biting back. You can't means-test it, or it stops being the thing.

Alex: And it gets worse. Compared with the targeted programmes a country already runs, a universal grant sends a far larger share of its money to households that aren't poor at all. The childless. The non-elderly. The non-disabled. The comfortably middle-income. You're spending vast sums giving money to people who don't need it — in order to preserve the principle that the floor is unconditional.

Sam: So you're burning enormous amounts of cash on people who are fine, just to keep the rule pure. And this is Krugman's trap from earlier, isn't it — the one you told me to hold onto.

Alex: This is Krugman's dilemma stated precisely, and nice callback. Make the grant adequate, and it's unaffordable. Make it affordable, and it's inadequate. There's no version that's both generous and cheap. And the political system has shown no appetite whatsoever for the taxes the generous version would actually require.

Sam: Okay, but that's a money problem, and money problems are at least the kind of thing societies sometimes solve. Is that the deepest issue with the evidence? Just the cost?

Alex: No — and this is the part I most want people to walk away with, because almost no reporting includes it. The deeper problem isn't cost. It's that every cheerful pilot result we've talked about carries an asterisk. Every one. The trials are, without exception, three things: dispersed, time-limited, and partial.

Sam: Unpack those three, because I think they sound minor and you're about t…